Background
The hippocampus is differentially affected in Alzheimer's disease neuropathologic change (ADNC) versus primary age-related tauopathy (PART), an amyloid-beta (Aβ)-independent tauopathy: the CA2/CA1 hyperphosphorylated tau (pTau)-ratio is higher in PART, which inversely correlates with Aβ-burden. However, as the aging brain often presents mixed rather than uniform pathologies, we questioned whether these distinct hippocampal pTau distributions persist into extreme ages and how hippocampal Aβ- and pTau-distributions correlate with cognition in centenarians.

Method
We quantified Aβ- (6F/3D) and pTau (AT8)-burdens across eight hippocampal and parahippocampal subregions in 112 centenarians (median age 104, IQR 102-105), alongside 11 AD (median age 84, IQR 72-86) and 7 PART cases for comparison (median age 88, IQR 78-92; Figure 1). We compared CA2/CA1-pTau-ratio in centenarians who met PART criteria (Thal phase ≤2, Braak stage I-IV; n = 49) with centenarians who met ADNC criteria (intermediate/high according to NIA-AA guidelines; Thal phase ≥3, Braak stage III-VI; n = 50). Cognitive performance was assessed using 13 neuropsychological tests shortly before brain donation (median 10 months, IQR5-14, n = 72). Robust linear regression models were used to associate subregional Aβ- and pTau-burdens with cognitive performance, while adjusting for age, sex, and education.

Result
In line with previous findings, CA2/CA1-pTau-ratios were higher in younger PART cases compared to AD patients (median 3.0, IQR 2.1-3.6, min-max 1.6-4.2 vs. median 1.2, IQR 0.9-1.4, min-max 0.8-1.4; p <0.001). Surprisingly, CA2/CA1-pTau-ratios in centenarians with PART were comparable to centenarians with ADNC (median 1.3, IQR 1.1-2.0, min-max 0.3-10.8 vs. median 1.2, IQR 1.0-1.8, min-max 0.2-6.2; p = 0.684). Accordingly, CA2/CA1-pTau-ratio in centenarians was unrelated to Aβ-burden, Thal phase or Braak stage. Higher Aβ- and pTau-burdens associated with lower cognition, though through different subregions: cognition associated with Aβ-burden in the hippocampus (CA4, CA3, CA2, CA1/subiculum), whereas pTau-burden in the parahippocampus (presubiculum, entorhinal cortex, fusiform gyrus) associated with cognition.

Conclusion
In the oldest-old, PART and ADNC are less distinguishable by determinants observed in younger individuals: centenarians with ADNC may show age-related Aβ accumulation alongside PART-like pTau patterns, while centenarians meeting PART criteria do not always show PART-like pTau patterns. However, hippocampal Aβ-burden and parahippocampal pTau-burden associate with cognitive decline, highlighting subregional-specific vulnerability to pathology-driven cognitive decline.

IMPORTANCE Older individuals without dementia often have amyloid-beta (Aβ) Thal phases similar to patients with Alzheimer disease (AD), suggesting that Aβ pathology may be a benign consequence of aging. OBJECTIVE To explore whether Aβ pathology in centenarians is associated with cognitive performance. DESIGN, SETTING, AND PARTICIPANTS This longitudinal cohort study used cross-sectional data on antemortem cognitive performance and postmortem neuropathology of participants in the Dutch 100-plus Study. Cognitive performance was measured a median of 10 (IQR, 3-13) months before postmortem brain donation. From January 2013 to July 2022, 1187 centenarians who self-reported being cognitively healthy, confirmed by proxy, were approached: 406 were included and 95 donated their brain. Centenarians were compared with patients with clinicopathologically confirmed AD from the Netherlands Brain Bank. Data were analyzed from June 2022 to October 2024. MAIN OUTCOMES AND MEASURES Aβ pathology was assessed with the Thal phase for Aβ progression and by determining quantitative Aβ loads (percentage positive area) in the frontal, parietal, temporal, and occipital neocortices, 3 parahippocampal, and 5 hippocampal subregions. Aβ pathology was associated with performance on 13 neuropsychological tests assessing memory, fluency, attention/processing speed, and executive functioning, as well as 4 measures of global cognition. RESULTS This study evaluated Aβ pathology in 95 centenarians (median age at brain donation, 103.5 [IQR, 102.3-104.7] years; 71 female [75%] and 24 male [25%]) and 38 patients with AD (median age, 84 [IQR, 78-90] years; 18 female [47%] and 20 male [53%]). Global cognition parameters were available for all 95 centenarians and complete cognitive assessment for 72 centenarians (76%). A fraction of the centenarians had no Aβ load (9 of 95 [9%]), most had low Aβ load (53 of 95 [56%]) and, despite high Thal phases, about one-third (33 of 95 [35%]) had high Aβ load comparable with patients with AD. Centenarians with no or low Aβ load had significantly higher cognitive performance than centenarians with high Aβ loads. Higher Aβ loads across all 4 neocortical regions, cornu ammonis 3, cornu ammonis 1/subiculum, and the entorhinal cortex specifically affected executive functioning. Interestingly, 5 resilient centenarians maintained high cognitive performance despite having high Aβ loads; they had significantly less tau pathology compared with centenarians with high Aβ loads and low cognitive performance. CONCLUSIONS AND RELEVANCE These results indicate that Aβ pathology is not a benign consequence of aging. Even in the oldest individuals, Aβ and tau pathology interaction was consistent with the amyloid cascade hypothesis.